JPH0314045A - History information management system for data base - Google Patents

Abstract

PURPOSE:To eliminate a need for a journal acquisition by providing a non- volatile memory with the data structure where update records point pri-update records in order from the latest record. CONSTITUTION:At the time of updating records, update system functions of data base system 3 and 8 hold update information of data base records and pri-update records on non-volatile memories 4 and 9 as data base records. The memory managing function discriminates whether it is necessary to sweep out records on non-volatile memories 4 and 9 to optical disk devices 5 and 10 or not based on the non-volatile memory use rate of records added in accordance with update of records, and they are swept out if it is necessary. At the time of the occurrence of trouble, the trouble recovering function updates the latest data base records based on pri-update records and deletes used pri- update records to restore the data base. Thus, it is unnecessary to independently get/manage the journal, and the degradation of performance or the complication of operation management accompanied with journal acquisition is prevented.

Description

[Detailed description of the invention]

[Industrial Application Field] The present invention relates to a database history information management method,
Especially in database systems, history information used for failure recovery and statistical analysis is not acquired separately from the database.
Database history information management method that treats update history information itself as a database [Prior art] In conventional database systems, information is centrally managed and used by multiple application programs. Therefore, it is necessary to maintain correct information at all times, including mutual relationships. However, unexpected situations often occur in the system, such as medium failure due to media destruction, system down due to power cut, or abnormal termination of a user program. When these failures occur, the information in the database becomes temporarily invalid, and the database acquires and manages journals so that this information can be correctly recovered. Furthermore, in order to ensure that the database can be properly recovered in the event of a failure, the journal separates the information before and after the update of the database, and is acquired and managed on a magnetic disk or magnetic tape device separately from the database. Furthermore, in order to maintain the consistency of the database information, the journal information is flushed to the medium before the database records are flushed to the database medium (magnetic disk device). Such a journal is used in five ways for database recovery in the event of a failure. That is, when a media failure occurs, a database operator recovers the database based on a database backup, which is a copy of the database contents acquired at a certain point in time, and a journal after the database has been updated. In addition, in cases other than media failure, the database system recovers the database based on the database at that point and the journal before database update at that point. On the other hand, with the spread of databases, companies that employ database systems have a desire to process the transition of updated information in chronological order and use it for business analysis and strategy planning. Note that related devices of this type include, for example, Japanese Patent Application Laid-Open No. 60-75950 and Japanese Patent Application Laid-Open No. 60-22043.
8, JP-A-61-844, JP-A-63-5
Publication No. 9640 and the like can be mentioned. [Problems to be Solved by the Invention] In the above-mentioned conventional technology, consideration is not given to a memory hierarchy using non-volatile memory such as a semiconductor disk device, and to the use of pre-database update information for purposes other than failures. In other words, since a method is adopted in which the database and journal are managed separately, there is a problem in that performance deteriorates due to journal acquisition and operational management becomes complicated. SUMMARY OF THE INVENTION An object of the present invention is to solve these problems and provide a database history information management system that does not require journal acquisition and can also handle time-series data. [Means for Solving the Problem] In order to achieve the above object, the database history information management method of the present invention uses a processing device, a nonvolatile memory that can be directly referenced, and a nonvolatile memory that can be directly referenced in a database system. an optical disc device, a means for accessing the optical disc device, an optical disc management table, and an optical disc buffer;
The non-volatile memory has a data structure in which updated records point to pre-updated records in order of newest, and data including updated records and pre-updated records are stored as database records, and a background job is used to update the database records. The feature is that it is flushed out to the optical disk device at regular intervals. [Operation] In the present invention, when updating a record, the update function of the database system updates the update information of the database record and the pre-update record as a database record.
Additionally, the memory management function determines whether or not it is necessary to flush records from non-volatile memory to the optical disk device based on the non-volatile memory usage rate of records added as records are updated. Then, if necessary, the optical disc device sweeps out the data. Further, the failure recovery function recovers the database when a failure occurs by updating the latest database record based on the pre-update record and deleting the used pre-update record. Further, the time series data processing function performs time series processing by using the data structure created by the time series index and update function. Therefore 1. There is no need to acquire/manage journals separately, and there is no performance deterioration or operational management complexity associated with journal acquisition. Furthermore, it is possible to process these data in time series. [Example] Hereinafter, an example of the present invention will be described in detail with reference to the drawings. FIG. 2 is an explanatory diagram showing a network system in an embodiment of the present invention. In FIG. 2, 1.6 is a processing unit, 2 and 7 are operating systems, 3 and 8 are database systems, and 4
, 9 are non-volatile memories, 5 and 10 are optical disk devices, 1
1 is a network. Further, the processing device 1 is equipped with an operating system 2 and a database system 3, and the database system 3 manages a database using a nonvolatile memory 4 and an optical disk device 5. Similarly, the processing device 6 has an operating system 7,
It is composed of a database system 8, a nonvolatile memory 9, and an optical disk drive 71110. By connecting these processing devices 'll and 6 through a network 11, a database system under a network environment is realized. Although the principle of the present invention can be realized by system identifier management in a distributed database using a wide area network or LAN, the basic principle is the same even with a single processing device, so from now on, we will focus on the 1H processing device. hand,
The configuration will be described using FIG. FIG. 3 is a configuration diagram of the database system in FIG. 2. In FIG. 3, 3a is a reference function, 3b is an update function, 3
C is a failure recovery function, 3d is a time series data processing function, 3e
3f is an exclusive control function, 3f is a memory management function, 3g is a nonvolatile memory access section, 3h is an optical disk device access section, 31 is a buffer, 3j is a time series index, 3 is a lock table, 31 is a memory management table, 3m is This is an optical disc management table. In this embodiment, the database system 3 of the processing device j is composed of three parts: a database function section, a table section used for each function, and an access section that accesses the nonvolatile memory 4 or the optical disk device 5. . This database function section utilizes a reference function 3a that refers to database records, an update function 3b that updates, adds, and deletes database records, and a failure recovery function 3C1 and time series index 3J that recovers the database when a failure occurs. Time series data processing function 3d that processes time series data at high speed, lock table 3
Also configured are an exclusive control function 3e that maintains data consistency of database records by using k, and a memory management function 3f that manages the location of database records by using a memory management table 31 and an optical disk management table 3m. Further, the access section includes a nonvolatile memory access section 3g that reads from or writes to the nonvolatile memory 4, and an optical disk device access section 3h that reads from or writes to the optical disk device 5 through the buffer 31. Further, the table section includes a time series index 3j, a lock table 3k that is referred to during exclusive control when updating data, a memory management table 31, and an optical disk management table 3m. Next, processing using these components will be described with reference to FIGS. 1 and 4. FIG. 1 is a flowchart showing history information management in one embodiment of the present invention, and FIG. 4 is a diagram showing an example of the data structure of a database record in one embodiment of the present invention. First, the case of a database record update request will be described. In this embodiment, a case is shown in which the database record shown in FIG.
O1). If the database record is on the optical disk device 5, the memory management function 3f uses the optical disk management table 3m to find the record position on the optical disk device 5, and instructs the optical disk device access section 3h to read the record. The optical disk device access section 3h reads the corresponding record from the optical disk device 5 into the buffer 31, and returns control to the memory management function 3f. The memory management function 3f provides a buffer 31 to the nonvolatile memory access unit 3g.
This command instructs the computer to transfer the information read into the nonvolatile memory 4 to the nonvolatile memory 4. Thereby, the memory management function 3f reflects these state changes in the memory management table 31, the optical disk management table 3m, and the time series index 3J (
102). Furthermore, after the distribution based on the processing request (lO3), the time series data processing function 3d acquires the corresponding record from the time series index j (104). Next, the update function 3b updates a record group (database records (1), (3) in FIG.
), (4)) to the exclusive control function 3e. As a result, the exclusive control function 3e locks these records in the lock table 3 (105) and returns control to the update system function 3b. The update system function 3b obtains free space from the memory management function 3f and creates a pre-update record (database record (2) in FIG. 4) (106).
2 updates the database record (database record (1) in FIG. 4), adds a pointer 41 to the previous record (2) (107, 108), and updates the time series index 3J using the time series data processing function. After that, the update system function 3b uses the exclusive control function 3e to release the lock information (110). Next, the update system function 3b tells the memory management function 3f to
A request is made to determine whether or not the contents of the non-volatile memory 4 need to be flushed to the optical disk device 5. The memory management function 3f determines whether flushing is necessary based on the usage rate of the nonvolatile memory 4 and the flushing time interval value (l l l). As a result, if it is necessary to purge the contents of the non-volatile memory 4, the contents of the non-volatile memory 4 are
and updates the optical disc management table 3m, memory management table 31, and time series index 3j (112 to 114). After this, the control is from the memory management function 3f to the update system function 3b.
and the process is completed. Next, the case of a database record search request will be described. In this embodiment, as in the case of an update request, the location of the requested record is found (101), the record is read from the optical disk device 5 into the nonvolatile memory 4 when necessary (102),
After the distribution based on the processing request (+03), the time series data processing function 3d quickly acquires the corresponding record from the time series index j using the hash function (116). Further, the reference function 3a displays the record obtained through these controls on the requesting terminal (117). Next, the case of a database recovery request will be described. In this embodiment, when the failure recovery function 3c obtains a group of records to be recovered from the time-series index 3J using time stamps, it processes n records at once instead of processing one record (1 Process n records
Database recovery is achieved using the same procedure as the database update process (101 to 103
．． 115).

【Effect of the invention】

According to the present invention, it is possible to recover a database without using a journal when a failure occurs, thereby eliminating the complexity of operational management of managing consistency between database backup acquisition timing and journal acquisition timing. Furthermore, since the journal is not acquired, it is possible to eliminate Ilo for journal acquisition in a journal acquisition device such as a display device, thereby improving the performance of the database system. Furthermore, since the database update information only needs to be held as a database record as an update result, the capacity of the recording medium required for the entire database system can be reduced. Furthermore, by having pre-database update information as a database record, it is also possible to perform time-series data processing, such as checking changes in period data.

[Brief explanation of drawings]

FIG. 1 is a flowchart showing history information management in an embodiment of the present invention, FIG. 2 is an explanatory diagram showing a network system on the actual flow side of the present invention, and FIG. 3 is a configuration diagram of the database system in FIG. 2. , FIG. 4 is a data configuration diagram of a database record in one embodiment of this embodiment. 1.6: Processing device, 2, 7: Operating system, 3, 8: Database system + 419 = non-volatile memory, 5, 10: Optical disk device. ll: network, 3a: reference function, 3b = update function, 3C: failure recovery function, 3d: time series data processing function,
3e: exclusive control function, 3f: memory management function, 3g = nonvolatile memory access section, 3h: optical disk device access section, 31: buffer, 3j: time series index, 2 lock table in 3, 31: memory management table, 3m =
Optical disc management table, 41-44: Pointer.

Claims (1)

[Claims] 1. In a database system, a processing device, a nonvolatile memory that can be directly referenced, a means for accessing the nonvolatile memory, an optical disk device, a means for accessing the optical disk device, an optical disk management table, The non-volatile memory has a data structure in which updated records point to pre-updated records in the order of newest, and data including updated records and pre-updated records is stored as database records and back A database history information management method characterized in that the database records are flushed to the optical disk device at regular intervals by a ground job.